Hi Stefan, I checked testweb before I posted my mail, to confirm the problem was going on for more than just my machine, and for how long. I guess the problem is related to gcc-4.2, or the Apple extensions. I hope the problem can be resolved somehow, since OSX is my favourite mobile development environment ;) Fabian On 16-02-2014 17:31:58 +0100, Stefan Manegold wrote:

Hi Fabian,

thanks for the configure fix and for the hint about the OSX (or its gcc version?) compilation problem.

For the compilation status (and testing) status, please see http://monetdb.cwi.nl/testweb/web/status.php (you might still recall that one ;-))

Indeed, also there the reported OSX compilation problem occurs: http://monetdb.cwi.nl/testweb/logs/50441:6e3415940261/GNU-Darwin-x86_64/make...

I guess, it is related to this changeset http://dev.monetdb.org/hg/MonetDB/rev/75da27800e54 but I'm not entirely sure.

We do still test on OSX and have not yet abandoned it as a "supported" platform (just like all other platforms we still test on), hence, I assume it's just that noone has had the time, yet, to look into the problem that appears to occur (1) "only" with the default branch and (2) "only" on OSX (and its gcc version).

Best, Stefan

On Feb 17, 2014, at 10:29, Hannes Mühleisen wrote:

Hi Fabian and list,

...

----- Original Message -----

...

Since I'm no longer really in the loop with MonetDB, I was wondering what the state of the default branch is supposed to be. I just fixed a configure error and now hit a redeclaration problem regarding gdk.h and gdk_private.h.

On 16.02.2014, at 17:31, Stefan Manegold wrote:

...

We do still test on OSX and have not yet abandoned it as a "supported" platform (just like all other platforms we still test on), hence, I assume it's just that noone has had the time, yet, to look into the problem that appears to occur (1) "only" with the default branch and (2) "only" on OSX (and its gcc version).

Problem has not appeared in my OSX setup, but I am not using gcc (clang instead...). So I would classify this as an OSX/gcc issue, not an OSX issue.

Hai Hannes, Since I upgrade my machine to Mavericks, I always get the error: libtool: link: gcc -g -Werror -Wall -Wextra -W -Werror-implicit-function-declaration -Wpointer-arith -Wdeclaration-after-statement -Wformat=2 -Wno-format-nonliteral -Winit-self -Winvalid-pch -Wmissing-declarations -Wmissing-format-attribute -Wmissing-prototypes -Wold-style-definition -Wpacked -Wunknown-pragmas -Wvariadic-macros -fstack-protector-all -Wstack-protector -Wmissing-include-dirs -D_REENTRANT -o .libs/tomograph tomograph.o -pthread -L/opt/local/lib ./.libs/libmcutil.a ../mapilib/.libs/libmapi.dylib /Users/jennie/scratch/monet-build/default/debug/common/stream/.libs/libstream.dylib ../../common/stream/.libs/libstream.dylib -lz -lbz2 -lssl -lcrypto -liconv -lcurl -pthread clang: error: argument unused during compilation: '-pthread' clang: error: argument unused during compilation: '-pthread' make[5]: *** [tomograph] Error 1 Any idea how to fix this? I'd really love to have monetdb running on my laptop again. Thanks Jennie

Best,

Hannes

_______________________________________________ developers-list mailing list developers-list@monetdb.org https://www.monetdb.org/mailman/listinfo/developers-list

what does your clang say with "-v" and "--dumpversion"? Stefan ----- Original Message -----

Hi,

I'm everything but a Mac or clang expert, but nightly testing seems to work fine both on our OSX machines (though not Mavericks, and using gcc from gentoo prefix) and on Fedora 20 using clang 3.3; cf., http://monetdb.cwi.nl/testweb/web/status.php?branch=default

Stefan

----- Original Message -----

...

On Feb 17, 2014, at 10:29, Hannes Mühleisen wrote:

...

Hi Fabian and list,

...

----- Original Message -----

...

Since I'm no longer really in the loop with MonetDB, I was wondering what the state of the default branch is supposed to be. I just fixed a configure error and now hit a redeclaration problem regarding gdk.h and gdk_private.h.

On 16.02.2014, at 17:31, Stefan Manegold wrote:

...

We do still test on OSX and have not yet abandoned it as a "supported" platform (just like all other platforms we still test on), hence, I assume it's just that noone has had the time, yet, to look into the problem that appears to occur (1) "only" with the default branch and (2) "only" on OSX (and its gcc version).

Problem has not appeared in my OSX setup, but I am not using gcc (clang instead...). So I would classify this as an OSX/gcc issue, not an OSX issue.

Hai Hannes,

Since I upgrade my machine to Mavericks, I always get the error:

libtool: link: gcc -g -Werror -Wall -Wextra -W -Werror-implicit-function-declaration -Wpointer-arith -Wdeclaration-after-statement -Wformat=2 -Wno-format-nonliteral -Winit-self -Winvalid-pch -Wmissing-declarations -Wmissing-format-attribute -Wmissing-prototypes -Wold-style-definition -Wpacked -Wunknown-pragmas -Wvariadic-macros -fstack-protector-all -Wstack-protector -Wmissing-include-dirs -D_REENTRANT -o .libs/tomograph tomograph.o -pthread -L/opt/local/lib ./.libs/libmcutil.a ../mapilib/.libs/libmapi.dylib /Users/jennie/scratch/monet-build/default/debug/common/stream/.libs/libstream.dylib ../../common/stream/.libs/libstream.dylib -lz -lbz2 -lssl -lcrypto -liconv -lcurl -pthread clang: error: argument unused during compilation: '-pthread' clang: error: argument unused during compilation: '-pthread' make[5]: *** [tomograph] Error 1

Any idea how to fix this? I'd really love to have monetdb running on my laptop again. Thanks

Jennie

...

Best,

Hannes

_______________________________________________ developers-list mailing list developers-list@monetdb.org https://www.monetdb.org/mailman/listinfo/developers-list

_______________________________________________ developers-list mailing list developers-list@monetdb.org https://www.monetdb.org/mailman/listinfo/developers-list

-- | Stefan.Manegold@CWI.nl | DB Architectures (DA) | | www.CWI.nl/~manegold/ | Science Park 123 (L321) | | +31 (0)20 592-4212 | 1098 XG Amsterdam (NL) |

_______________________________________________ developers-list mailing list developers-list@monetdb.org https://www.monetdb.org/mailman/listinfo/developers-list

-- | Stefan.Manegold@CWI.nl | DB Architectures (DA) | | www.CWI.nl/~manegold/ | Science Park 123 (L321) | | +31 (0)20 592-4212 | 1098 XG Amsterdam (NL) |

Thanks Hannes! ... I'd prefer, though, to understand why it works with nightly testing, but fails for Jennie (and you?), respectively how to distinguish working from failing setups, and then "fix" our configure (in a "conservative" manner), such that all our environments do work ... Best, Stefan ----- Original Message -----

Hello Jennie (and Stefan),

On 04.03.2014, at 11:45, Ying Zhang wrote:

...

clang: error: argument unused during compilation: '-pthread’

If you set

export CFLAGS="-Qunused-arguments”

it should compile.

Best,

Hannes

_______________________________________________ developers-list mailing list developers-list@monetdb.org https://www.monetdb.org/mailman/listinfo/developers-list

-- | Stefan.Manegold@CWI.nl | DB Architectures (DA) | | www.CWI.nl/~manegold/ | Science Park 123 (L321) | | +31 (0)20 592-4212 | 1098 XG Amsterdam (NL) |

@Hannas: the CFLAGS helps, thanks!

@Stefan: this is because, as of Mavericks, the command "gcc" is not really GCC any more, it's just a copy of Clang: $ gcc --version Configured with: --prefix=/Applications/Xcode.app/Contents/Developer/usr --with-gxx-include-dir=/usr/include/c++/4.2.1 Apple LLVM version 5.0 (clang-500.2.79) (based on LLVM 3.3svn) Target: x86_64-apple-darwin13.1.0 Thread model: posix $ clang -v Apple LLVM version 5.0 (clang-500.2.79) (based on LLVM 3.3svn) Target: x86_64-apple-darwin13.1.0 Thread model: posix

thanks. that's what I wanted to know, i.e., this is not only not gcc, but also different from the clang 3.3 that's available on Linux. Apparently, the latter (still?) supports -pthread while the Apple one does no(t| longer).

which doesn't support GCC-only flags. Can we some how update our configure script or something to deal with this in a more elegant way than having to set "export CFLAGS..." each time?

sure. See e.g. what we do already do for clang, icc, and certain (old) versions of gcc. But I propose to check carefully and understand why configure itself or our checks in our configure.ag does not (yet) detect that Apple clang 5.0 does not support -pthread (the way it is currently used). Stefan

Jennie

On Mar 04, 2014, at 13:29, Stefan Manegold wrote:

...

Thanks Hannes!

... I'd prefer, though, to understand why it works with nightly testing, but fails for Jennie (and you?), respectively how to distinguish working from failing setups, and then "fix" our configure (in a "conservative" manner), such that all our environments do work ...

Best, Stefan

----- Original Message -----

...

Hello Jennie (and Stefan),

On 04.03.2014, at 11:45, Ying Zhang wrote:

...

clang: error: argument unused during compilation: '-pthread’

If you set

export CFLAGS="-Qunused-arguments”

it should compile.

Best,

Hannes

_______________________________________________ developers-list mailing list developers-list@monetdb.org https://www.monetdb.org/mailman/listinfo/developers-list

-- | Stefan.Manegold@CWI.nl | DB Architectures (DA) | | www.CWI.nl/~manegold/ | Science Park 123 (L321) | | +31 (0)20 592-4212 | 1098 XG Amsterdam (NL) |

_______________________________________________ developers-list mailing list developers-list@monetdb.org https://www.monetdb.org/mailman/listinfo/developers-list

_______________________________________________ developers-list mailing list developers-list@monetdb.org https://www.monetdb.org/mailman/listinfo/developers-list

-- | Stefan.Manegold@CWI.nl | DB Architectures (DA) | | www.CWI.nl/~manegold/ | Science Park 123 (L321) | | +31 (0)20 592-4212 | 1098 XG Amsterdam (NL) |

Thanks, Fabian! That's basically what I meant with "understanding what's going on" ;-) I'll look into this, once I receive input on how gcc/clang reports its version/identity on Mavericks (see my previous mail). Stefan ----- Original Message -----

On 04-03-2014 20:26:28 +0100, Stefan Manegold wrote:

...

thanks. that's what I wanted to know, i.e., this is not only not gcc, but also different from the clang 3.3 that's available on Linux. Apparently, the latter (still?) supports -pthread while the Apple one does no(t| longer).

No, it just became more strict. It is warning that -pthread doesn't do anything during compilation phase. You can argue this is a clang bug, since -pthread is a compiler switch that says: "do whatever you need to do to get compilation of code against pthread working". In the Darwin case (and also Linux) it doesn't have to do anything.

...

...

which doesn't support GCC-only flags. Can we some how update our configure script or something to deal with this in a more elegant way than having to set "export CFLAGS..." each time?

sure. See e.g. what we do already do for clang, icc, and certain (old) versions of gcc. But I propose to check carefully and understand why configure itself or our checks in our configure.ag does not (yet) detect that Apple clang 5.0 does not support -pthread (the way it is currently used).

Again, it does support that flag. You just can't compile with --enable-strict. I suggest you change the configure trickery to disable warning about unused flags for Clang 5.

€0.02

Fabian

_______________________________________________ developers-list mailing list developers-list@monetdb.org https://www.monetdb.org/mailman/listinfo/developers-list

-- | Stefan.Manegold@CWI.nl | DB Architectures (DA) | | www.CWI.nl/~manegold/ | Science Park 123 (L321) | | +31 (0)20 592-4212 | 1098 XG Amsterdam (NL) |

On 04-03-2014 20:55:42 +0100, Stefan Manegold wrote:

Jennie (or someone else running Mavericks),

could you please provide me with the output (stdout & stderr) of these four commands:

gcc -dumpversion

gcc -v

clang -dumpversion

clang -v

on Mavericks?

[nut:gentoo-x86/mail-client/mutt] % /usr/bin/gcc -dumpversion 4.2.1 [nut:gentoo-x86/mail-client/mutt] % /usr/bin/gcc -V clang: error: argument to '-V' is missing (expected 1 value) clang: error: no input files [nut:gentoo-x86/mail-client/mutt] % /usr/bin/gcc --version Configured with: --prefix=/Library/Developer/CommandLineTools/usr --with-gxx-include-dir=/usr/include/c++/4.2.1 Apple LLVM version 5.0 (clang-500.2.79) (based on LLVM 3.3svn) Target: x86_64-apple-darwin13.1.0 Thread model: posix [nut:gentoo-x86/mail-client/mutt] % /usr/bin/gcc -v Configured with: --prefix=/Library/Developer/CommandLineTools/usr --with-gxx-include-dir=/usr/include/c++/4.2.1 Apple LLVM version 5.0 (clang-500.2.79) (based on LLVM 3.3svn) Target: x86_64-apple-darwin13.1.0 Thread model: posix [nut:gentoo-x86/mail-client/mutt] % /usr/bin/clang -dumpversion 4.2.1 [nut:gentoo-x86/mail-client/mutt] % /usr/bin/clang -v Apple LLVM version 5.0 (clang-500.2.79) (based on LLVM 3.3svn) Target: x86_64-apple-darwin13.1.0 Thread model: posix [nut:gentoo-x86/mail-client/mutt] % sw_vers ProductName: Mac OS X ProductVersion: 10.9.2 BuildVersion: 13C64

Jennie, I just checked in a configure fix (http://dev.monetdb.org/hg/MonetDB/rev/36afc568f849) (http://dev.monetdb.org/hg/MonetDB/rev/a06afa30eeea) Could you please test whether that works on Mac OSX Mavericks? I.e., try to build MonetDB from scratch, preferably configured with --enable-optimize and --enable-strict (and also keep an eye on the configure output) Thanks! Stefan -- | Stefan.Manegold@CWI.nl | DB Architectures (DA) | | www.CWI.nl/~manegold/ | Science Park 123 (L321) | | +31 (0)20 592-4212 | 1098 XG Amsterdam (NL) |

On 05-03-2014 09:01:21 +0100, Hannes Mühleisen wrote:

Here is the output from my box, looks consistent! Fabian, I think you accidentally used -V on gcc, not -v.

I used all of them, on purpose :)

On 17-02-2014 08:28:15 +0100, Sjoerd Mullender wrote:

Can you retry the compilation? I had moved some struct definitions to gdk_private.h, and just left pointer-to-struct in gdk.h. But the double typedef of the same struct (one with, one without content) may have caused the complaint.

That seems to have helped, thanks! Fabian